What Is Particle Swarm Optimization (PSO)?

Particle swarm optimization (PSO) is a population-based stochastic method that helps with optimization problems. It is modeled after natural processes, such as the flocking of birds or the movement of schools of fish. PSO was first developed in the early 1990s by Dr. James Kennedy and Dr. Eberhart Zitzler at the University of Colorado. They were trying to solve their university's problem of finding the shortest route across campus for students and staff to take between buildings. As it turns out, this type of problem is called route discovery, which many companies face daily. The PSO algorithm starts with a randomly generated population of particles (or "birds," if you will). Each particle illustrates a potential solution to the overall problem at hand. For example, in our example above, these might be campus roads or paths students can walk from one building to another. The particles have value scores associated with them. These values represent how well each particle matches what we're looking for (i.e., how good they are at helping us find a particular path). The value scores are updated based on each particle's performance against other particles during interactions. The idea behind particle swarm optimization is simple: you want to know what's happening in a peer-to-peer network, but you want to spend only some days watching it. So you give your computer a bunch of birds and let them fly around the network, reporting to you on their findings. It's like that old game where you put out a bunch of colored balls and try to get them in the correct order. Instead of balls, you have birds…and they tell you where they are instead of putting them in order. Then you can use those reports to figure out what's happening in the network or predict how things will go!